Quick Summary
Automation relies on the `alert_rules` and `cron` daemons. The AI layer (`ai assistant`) uses grounded embeddings to provide accurate summaries of the `jobs` table, bypassing manual search.
Configuring AI Grounding
The AI assistant generates answers based exclusively on data in your tenant database. It does not hallucinate generalized advice.
- Navigate to System Setup > AI Settings (
ai settings). - Input the API key. The system encrypts this before inserting it into the
ai_settingstable. - Open the AI Assistant chat interface (
ai assistant chat). - Submit a query (e.g., "What jobs are overdue?"). The backend executes SQL matching your query, tokenizes the records, and streams the inference result back.
Defining Workflow Alert Rules
Workflow rules listen for state mutations (e.g., a job status changing to "Complete") and trigger subsequent actions.
- Navigate to Automation > Alert Rules and click + Add Alert Rule (
alert rule add new). - Set the Trigger Event (e.g., Job Status Changed).
- Set the Condition (e.g., Status equals "Invoiced").
- Set the Action (e.g., Notify Subcontractor).
- Click Save. When
job update processruns, it will evaluate this rule array and dispatch notifications if the condition evaluates to true.